seurat compatibility

[Elena983]

How can the SpotWeeper be used with the Seurat obj? Otherwise I need to transit it to spe and then back

[MicTott]

Hi @Elena983!

Your message came at a good time. I actually just pushed Seurat compatibility in the latestdevelopment version (v1.5.0) for the localOutliers function. I have not extensively tested this functionality yet though.

If you want to give it a try, you can install the latest development version using:

if (!require("devtools")) install.packages("devtools")
remotes::install_github("MicTott/SpotSweeper")

Let me know if you try it out and how it goes. Reporting any potential issues here would go along way to full Seurat compatibility. I will update here when I find the time to incorporate compatibility to other functions.

Best,
Michael

[Elena983]

I will try, but we can use it in production mode eventually, only after a stable version is released. Look forward!

[Elena983]

Hi @MicTott

Can you please provide me with an example for using the seurat object?
Thanks)

[Elena983]

So, I tested the seurat obj.

works well for the 1st called metric, but on the next one, it crushes, saying

Error in subset.Seurat(x, cells = cells_to_keep) :
No cells left in the default assay, please change the default assay

I researched the error.

localOutliers() subsetting can wipe out assays due to a misaligned logical mask in subsetSpatialObject()

The root cause appears to be subsetting due to a misaligned logical mask in the internal subsetSpatialObject() call. The logical vector is constructed from getMetadata(spe), which is not guaranteed to be aligned with Cells(spe), leading to the selection of 0 cells and triggering Seurat’s misleading error.

getMetadata(spe) returns a frame where rownames(getMetadata(spe)) do not equal Cells(spe).

Using a logical mask derived from that unaligned frame causes subset() to select 0 cells.

The error message is misleading; the root cause is empty selection, not a priori empty assays.

I guess you can at least always derive metadata aligned to cells for Seurat via spe[[]] and subset by explicit cell names, not by a logical mask.

[domenico-somma]

Hi,
I am also interested to this, and I got the same error if you use the 2 command one after another, but not if they are separated:

#Load Xenium data as Seurat
test = LoadXenium("/mypath", segmentations = "cell")
#Log normalize
test <- NormalizeData(test)

#This command on the original object works:
test2 = localOutliers(test, samples = "orig.ident", metric = "nCount_Xenium")
table(test2$nCount_Xenium_outliers)
 FALSE   TRUE 
156732    955 

#This command also works on the original object:
test3 = localOutliers(test, samples = "orig.ident", metric = "nFeature_Xenium")
table(test3$nFeature_Xenium_outliers)
 FALSE   TRUE 
156518   1169 

#But they gave error if I generated them on SpotSweeper output:
test2 = localOutliers(test, samples = "orig.ident", metric = "nCount_Xenium")
table(test2$nCount_Xenium_outliers)
 FALSE   TRUE 
156732    955 

#This command also works on the original object:
test3 = localOutliers(test2, samples = "orig.ident", metric = "nFeature_Xenium")

Warning: Cannot find cells provided
Xenium assay doesn't leave any cells, so it is removed
Error in subset.Seurat(x, cells = cells_to_keep) : 
  No cells left in the default assay, please change the default assay

> packageVersion("Seurat")
[1] ‘5.3.0’
> packageVersion("SpotSweeper")
[1] ‘1.5.0’

[MicTott]

Hi @Elena983 and @domenico-somma, thanks for the feedback! I see the error now and will try to find time before the end of the year to push a fix through. It's a busy time for me, so thanks for your patience!

Best,
Michael